EMIS/ELibM Electronic Journals

Outdated Archival Version

These pages are not updated anymore. They reflect the state of 20 August 2005. For the current production of this journal, please refer to http://www.math.psu.edu/era/.

%_ **************************************************************************
%_ * The TeX source for AMS journal articles is the publishers TeX code     *
%_ * which may contain special commands defined for the AMS production      *
%_ * environment.  Therefore, it may not be possible to process these files *
%_ * through TeX without errors.  To display a typeset version of a journal *
%_ * article easily, we suggest that you either view the HTML version or    *
%_ * retrieve the article in DVI, PostScript, or PDF format.                *
%_ **************************************************************************
%  Author Package
%% Translation via Omnimark script a2l, March 22, 1999 (all in one day!)

%% Declarations:


\newtheorem*{definition1}{Step I}
\newtheorem*{definition2}{Step II}
\newtheorem*{definition3}{Step III}
\newtheorem*{definition4}{Step IV}
\newtheorem*{definition5}{Step V}

%% User definitions:


{On Bojarski's index formula for nonsmooth interfaces}
\author{Marius Mitrea}
\address{Department of Mathematics,
University of Missouri-Columbia,
Columbia, MO 65211}
\subjclass{Primary 58G10, 42B20; Secondary 34L40, 30D55}
\thanks{Partially supported by NSF}
\dateposted{April 6, 1999}
\PII{S 1079-6762(99)00060-8}
\copyrightinfo{1999}{American Mathematical Society}
\commby{Stuart Antman}
\date{December 02, 1998}
Let $D$ be a Dirac type operator on a compact manifold ${\mathcal{M}}$
and let $\Sigma $ be a Lipschitz submanifold of codimension one 
${\mathcal{M}}$ into two Lipschitz domains $\Omega _{\pm }$. Also, let
${\mathcal{H}}^{p}_{\pm }(\Sigma ,D)$ be the traces on $\Sigma $ of the
($L^{p}$-style) Hardy spaces associated with $D$ in $\Omega _{\pm }$.
Then $({\mathcal{H}}^{p}_{-}(\Sigma ,D),{\mathcal{H}}^{p}_{+}(\Sigma 
,D))$ is
a Fredholm pair of subspaces for $L^{p}(\Sigma )$ (in Kato's sense) whose
index is the same as the index of the Dirac operator $D$ considered
on the whole manifold ${\mathcal{M}}$.

\section{Introduction and statement of results}Let $\Sigma $ be 
a rectifiable curve in ${\mathbb{R}}^{2}\equiv {\mathbb{C}}$
which divides the plane into two domains, denoted by $\Omega _{\pm }$.
Corresponding to these, introduce the Hardy spaces on $\Sigma $,
\begin{equation}\label{eq:1.1}{\mathcal{H}}^{p}_{\pm }(\Sigma ):=\{u|_{\Sigma 
};\,\bar {\partial }u=0\,\,
\text{in}\,\,\Omega _{\pm },\,\,{\mathcal{N}}(u)\in L^{p}(\Sigma )\},
\qquad 1m:=\text{dim}\,{\mathcal{M}}$.
Our aim is to investigate the analogue of \eqref{eq:1.2} when $\Sigma $
is a Lipschitz submanifold of codimension one of ${\mathcal{M}}$ and
$D:{\mathcal{E}}\to {\mathcal{F}}$ is a first-order, elliptic 
operator whose coefficients have a limited amount of smoothness.
Specifically, we assume that
\begin{equation}\label{eq:1.3}\text{the top coefficients of}\,\,D\,\,\text{belong 
\begin{equation}\label{eq:1.4}\text{the coefficients of the zero order part 
\text{belong to}\,\,H^{1,r},
for some $r>m$. Under the current assumptions, $D^{*}$, the adjoint of 
also satisfies \eqref{eq:1.3}--\eqref{eq:1.4}.

The major analytical hypothesis we make on the differential operator
is that
\begin{equation}\label{eq:1.5}D\,\,\text{and}\,\,D^{*}\,\,\text{have the unique 
Recall that $D$ is said to have the unique continuation property
if the following holds: $u\in H^{1,2}({\mathcal{M}},{\mathcal{E}})$
and $Du=0$ implies that either $u\equiv 0$ or
$\text{supp}\,u={\mathcal{M}}$. The hypothesis \eqref{eq:1.5} is natural
inasmuch as it is automatically satisfied (for $r=\infty $) by
Dirac type operators; cf. \cite{Ar}, \cite{Co}.

Consider now a Lipschitz subdomain $\Omega $ of ${\mathcal{M}}$ and
set $\Sigma :=\partial \Omega $, $\Omega _{+}:=\Omega $,
$\Omega _{-}:={\mathcal{M}}\setminus \bar {\Omega }$. Also, for 
$10$. Hence, the integral operator $K:=C_{+}-C_{-}$ is compact
from $L^{p}(\Sigma ,{\mathcal{E}})$ into itself, $10.
These are proved by implementing a symbol decomposition to the
effect that $D=D^{\#}+D^{b}+B$ with
\begin{equation}\label{eq:2.16}D^{\#}\in OPC^{\infty }S^{1}_{1,\delta },\quad 
D^{b}\in OPC^{1}S^{1-\delta }_{1,\delta },\quad B \in L^{\infty 
for some $0<\delta <1$. Now \eqref{eq:2.14}--\eqref{eq:2.15} follow from the existence
of a parametrix for $D^{\#}$ and mapping properties for
pseudodifferential operators whose symbols have a limited amount
of smoothness. See, e.g., \cite{Ta} for references and a general
discussion of such issues.

Now, if $f\in {\mathcal{H}}^{p}_{-}(\Sigma ,D)\cap {\mathcal{H}}^{p}_{+
}(\Sigma ,D)$,
then there exist (unique, by \eqref{eq:2.6}) functions
$u_{\pm }\in {\mathcal{H}}^{p}(\Omega _{\pm },D)$ so that
$u_{+}|_{\Sigma }=f=u_{-}|_{\Sigma }$. Set $u:=u_{-}$ in $\Omega _{-}$ 
$u:=u_{+}$ in $\Omega _{+}$, $u\in L^{p}({\mathcal{M}},{\mathcal{E}})$ 
so that
$u\in \text{Ker}\,\left (D:H^{1,p}({\mathcal{M}},{\mathcal{E}})\to 
L^{p}({\mathcal{M}},{\mathcal{F}})\right )$. Consequently, the (just 
assignment $f\mapsto u$ is linear, well defined and, by \eqref{eq:2.6},
one-to-one from ${\mathcal{H}}^{p}_{-}(\Sigma ,D)\cap 
{\mathcal{H}}^{p}_{+}(\Sigma ,D)$
into $\text{Ker}\,\left (D:H^{1,p}({\mathcal{M}},{\mathcal{E}})\to 
L^{p}({\mathcal{M}},{\mathcal{F}})\right )$.
Invoking \eqref{eq:2.15}, it is clear that this is also onto. Hence,
\begin{equation}\label{eq:2.17}\text{dim}\left ({\mathcal{H}}^{p}_{-}(\Sigma ,D)
\cap {\mathcal{H}}^{p}_{+}(\Sigma ,D)\right )
=\text{dim}\,\text{Ker}\left (D:H^{1,p}({\mathcal{M}},{\mathcal{E}})
\to L^{p}({\mathcal{M}},{\mathcal{F}})\right )<\infty .
At this point, the first part in Theorem \ref{thm:1} follows. There remain
\eqref{eq:1.9} and \eqref{eq:1.10}, which require refining further the analysis carried out
so far. This makes the object of the next couple of steps.
\begin{definition4} Let $[...]^{\circ }$ stand for the annihilator of
$[...]$ under the pairing $(u,v)=\int _{\Sigma }\langle u,v^{c}\rangle 
We shall prove that the mapping
\begin{equation}\label{eq:2.18}\Phi :{\mathcal{H}}^{p}_{\pm }(\Sigma ,D^{*})\to \left 
[{\mathcal{H}}^{q}_{\pm }(\Sigma ,D)\right ]^{\circ },\,\,
\Phi (u):=i\sigma (D^{*};\nu )u,\quad 1/p+1/q=1,
is an isomorphism for any $1